
    9|hR                       d Z ddlmZ ddlmZmZmZmZmZm	Z	 ddl
mZ ddlmZmZ ddlmZ ddlmZ ddlmZ dd	lmZ dd
lmZ ddlmZ ddlmZ ddlmZ ddlm Z   eddd       G d de              Z! G d deee"e"f            Z#y)z+Base classes for LLM-powered router chains.    )annotations)AnyDictListOptionalTypecast)
deprecated)AsyncCallbackManagerForChainRunCallbackManagerForChainRun)OutputParserException)BaseLanguageModel)BaseOutputParser)BasePromptTemplate)parse_and_check_json_markdown)model_validator)SelfLLMChain)RouterChainz0.2.12z1.0zUse RunnableLambda to select from multiple prompt templates. See example in API reference: https://api.python.langchain.com/en/latest/chains/langchain.chains.router.llm_router.LLMRouterChain.html)sinceremovalmessagec                       e Zd ZU dZded<   	  ed      dd       Zedd       Zd fdZ		 d	 	 	 	 	 dd	Z
	 d	 	 	 	 	 dd
Ze	 	 	 	 	 	 	 	 dd       Z xZS )LLMRouterChaina
	  A router chain that uses an LLM chain to perform routing.

    This class is deprecated. See below for a replacement, which offers several
    benefits, including streaming and batch support.

    Below is an example implementation:

        .. code-block:: python

            from operator import itemgetter
            from typing import Literal
            from typing_extensions import TypedDict

            from langchain_core.output_parsers import StrOutputParser
            from langchain_core.prompts import ChatPromptTemplate
            from langchain_core.runnables import RunnableLambda, RunnablePassthrough
            from langchain_openai import ChatOpenAI

            llm = ChatOpenAI(model="gpt-4o-mini")

            prompt_1 = ChatPromptTemplate.from_messages(
                [
                    ("system", "You are an expert on animals."),
                    ("human", "{query}"),
                ]
            )
            prompt_2 = ChatPromptTemplate.from_messages(
                [
                    ("system", "You are an expert on vegetables."),
                    ("human", "{query}"),
                ]
            )

            chain_1 = prompt_1 | llm | StrOutputParser()
            chain_2 = prompt_2 | llm | StrOutputParser()

            route_system = "Route the user's query to either the animal or vegetable expert."
            route_prompt = ChatPromptTemplate.from_messages(
                [
                    ("system", route_system),
                    ("human", "{query}"),
                ]
            )


            class RouteQuery(TypedDict):
                """Route query to destination."""
                destination: Literal["animal", "vegetable"]


            route_chain = (
                route_prompt
                | llm.with_structured_output(RouteQuery)
                | itemgetter("destination")
            )

            chain = {
                "destination": route_chain,  # "animal" or "vegetable"
                "query": lambda x: x["query"],  # pass through input query
            } | RunnableLambda(
                # if animal, chain_1. otherwise, chain_2.
                lambda x: chain_1 if x["destination"] == "animal" else chain_2,
            )

            chain.invoke({"query": "what color are carrots"})
    r   	llm_chainafter)modec                `    | j                   j                  }|j                  t        d      | S )NzLLMRouterChain requires base llm_chain prompt to have an output parser that converts LLM text output to a dictionary with keys 'destination' and 'next_inputs'. Received a prompt with no output parser.)r   promptoutput_parser
ValueError)selfr    s     a/var/www/html/test/engine/venv/lib/python3.12/site-packages/langchain/chains/router/llm_router.pyvalidate_promptzLLMRouterChain.validate_prompth   s6    &&'      c                .    | j                   j                  S )zTWill be whatever keys the LLM chain prompt expects.

        :meta private:
        )r   
input_keys)r#   s    r$   r(   zLLMRouterChain.input_keyst   s     ~~(((r&   c                V    t         |   |       t        |d   t              st        y )Nnext_inputs)super_validate_outputs
isinstancedictr"   )r#   outputs	__class__s     r$   r,   z LLMRouterChain._validate_outputs|   s*    !'*'-0$7 8r&   c                &   |xs t        j                         }|j                         } | j                  j                  dd|i|}t        t        t        t        f   | j                  j                  j                  j                  |            }|S N	callbacks )r   get_noop_manager	get_childr   predictr	   r   strr   r    r!   parse)r#   inputsrun_manager_run_managerr3   
predictionoutputs          r$   _callzLLMRouterChain._call   s~    
 #S&@&Q&Q&S **,	+T^^++JiJ6J
cNNN!!//55jA
 r&   c                   K   |xs t        j                         }|j                         }t        t        t
        t        f    | j                  j                  dd|i| d {         }|S 7 
wr2   )	r   r5   r6   r	   r   r8   r   r   apredict_and_parse)r#   r:   r;   r<   r3   r>   s         r$   _acallzLLMRouterChain._acall   sk     
 #S&@&Q&Q&S **,	cN3$..33RiR6RR
  Ss   A!A0#A.
$A0c                0    t        ||      } | dd|i|S )zConvenience constructor.)llmr    r   r4   r   )clsrD   r    kwargsr   s        r$   from_llmzLLMRouterChain.from_llm   s#    
 V4	1Y1&11r&   )returnr   )rH   z	List[str])r/   Dict[str, Any]rH   None)N)r:   rI   r;   z$Optional[CallbackManagerForChainRun]rH   rI   )r:   rI   r;   z)Optional[AsyncCallbackManagerForChainRun]rH   rI   )rD   r   r    r   rF   r   rH   r   )__name__
__module____qualname____doc____annotations__r   r%   propertyr(   r,   r?   rB   classmethodrG   __classcell__)r0   s   @r$   r   r      s    AF +'"	 #	 ) ) =A : 
	$ BF ? 
	 2#2-?2KN2	2 2r&   r   c                  D    e Zd ZU dZdZded<   eZded<   dZded<   dd	Z	y
)RouterOutputParserz<Parser for output of router chain in the multi-prompt chain.DEFAULTr8   default_destinationr   next_inputs_typeinputnext_inputs_inner_keyc                   	 ddg}t        ||      }t        |d   t              st        d      t        |d   | j                        st        d| j                   d      | j
                  |d   i|d<   |d   j                         j                         | j                  j                         k(  rd |d<   |S |d   j                         |d<   |S # t        $ r}t        d| d|       d }~ww xY w)Ndestinationr*   z&Expected 'destination' to be a string.zExpected 'next_inputs' to be .zParsing text
z
 raised following error:
)r   r-   r8   r"   rW   rY   striplowerrV   	Exceptionr   )r#   textexpected_keysparsedes        r$   r9   zRouterOutputParser.parse   s   	*M:M24GFf]3S9 !IJJf]3T5J5JK 3D4I4I3J!L  &*%?%?AV$WF=!}%++-335++1134 )-}% M )/}(=(C(C(E}%M 	' &B1#F 	s   B6C 9C 	C0C++C0N)r`   r8   rH   rI   )
rK   rL   rM   rN   rV   rO   r8   rW   rY   r9   r4   r&   r$   rT   rT      s+    F(( d !(3(r&   rT   N)$rN   
__future__r   typingr   r   r   r   r   r	   langchain_core._apir
   langchain_core.callbacksr   r   langchain_core.exceptionsr   langchain_core.language_modelsr   langchain_core.output_parsersr   langchain_core.promptsr   langchain_core.utils.jsonr   pydanticr   typing_extensionsr   langchain.chainsr   langchain.chains.router.baser   r   r8   rT   r4   r&   r$   <module>rq      s    1 " 8 8 * < < : 5 C $ " % 4 
	s	B2[ B2B2J)$sCx.9 r&   